Developing the Luciano Floridi (LuFlot) Bot: An Accessible AI Chatbot Trained on a Philosopher’s Manuscripts
摘要
This paper presents the Luciano Floridi Bot (LuFlot), the first freely and publicly accessible AI chatbot trained on a corpus of literature by a living philosopher, designed to broaden access to insights into AI ethics, digital ethics, the philosophy of information, and the philosophy of technology. Built using OpenAI’s Assistants API and implementing Retrieval-Augmented Generation (RAG) technology, LuFlot leverages the manuscripts of Luciano Floridi’s works to provide cited, academically rigorous responses to questions about digital ethics and philosophy, while maintaining accessibility for diverse audiences. The system employs advanced prompt engineering techniques, including Chain of Thought and Few-Shot prompting, and offers multilingual capabilities and adaptable explanation levels. Unlike existing philosophical chatbots, LuFlot offers free access, direct citations, and customizable interaction depth tailored to each user’s query. While the system demonstrates promising applications in education and research, particularly for researchers in philosophy, its limitations include the potential for reductionism and the need for critical evaluation of its outputs. This implementation demonstrates how pedagogically optimized AI systems can effectively mediate between complex academic thought and public understanding while maintaining scholarly integrity through proper attribution.